THE EXTREME VALUES OF DAILY PRECIPITATION IN JAPAN (2 ND REPORT)

Abstract

In the previous study (Geographical Review of Japan. Vol. 31, No. 2, pp. 86_??_94, 1958), the author estimated the annual and seasonal maximum daily precipitations for 5, 10, 25, 50, and 100-year return periods by applying the Gumbel's theory. Using the estimated data, distribution maps of extreme values were constructed. But it was pointed out that the Gumbel's theory contained some questions. One of them is the fitness to actual distribution of extreme values. According to the works by Brooks, C. F. P. and N. Carruthers (1953), the extreme values of air temperature are liable to be overestimated and those of precipition understimated, if the Gumbel's theory is used. Jenkinson, A. F. (1955) presented an improved theory for the estimation of extreme values. This shows better fitness and the process of estimation is not so difficult. Hence the author reestimated the extreme values according to this theory for several return periods in Japan. The results at a number of stations were added to the previous works. Regional distribution of the extreme values generally shows similar patterns to the previous maps (Fig. 2). Details of distribution, however, have some different features from the previous maps. The following facts were made clear in this paper:1) The June-July season: heavy rainfall areas appear in western Kyushu, southeaster part of Kii and entire part of the Izu peninsula.2) The August-November season: heavy rainfall areas appear in the eastern Kyushu, southeastern Kii peninsula, and northern and western Kanto.3) The December-May season: an area with scanty rainfall appears in the coastal regions of Seto Inland Sea.4) Year: heavy rainfall areas appear in southeastern and western Kyushu and western Kanto, and an area with scanty rainfall inn the coastal regions of Seto Inland Sea.The author compared the estimated extreme values of the June-July season with those of the August November season. Heavy rainfalls in the June-July season are mainly caused frontal activities (Bai-u), and those in the August-November season are by typhoons. At stations where the estimated values in the June-July season are larger than those in the August-November season, fronal rainfall is distinctly heavier than typhoon rainfall. On the contrary, at stations where the estimated values in the August-November season are larger than those in June-July, typhoon rainfall is heavier than frontal rainfall. The author calculated the ratio of the extreme values in the June-July season to those in the August-November season for each station. At the stations with the ratio exceeding 1. 0, frontal rainfall is predominant, whereas typhoon rainfall is abundant at the stations with the ratio below 1.0. Areas with the ratio exceeding 1.0 are concentrated in the following districts (Bai-u areas): 1) Northern and western Kyushu, 2) Chugoku (except eastern San'in), 3) Western Shikoku, 4) Middle Kinki, 5) Hokuriku. Areas where the stations with the ratio less than 1.0 are concentrated are the following districts (typhoon areas): 1) Eastern Kyukhu, 2) Eastern Shikoku, 3) Northern and southwestern Kinki, 4) Western Kanto and Izu islands, 5) Eastern Ou, 6) Northeartern Hokkaido.Based on these facts, Bai-u and Typhoon areas can be distinguished in Japan. Lastly, the author made clear the regionalities that a found in the secular changes of maximum . daily precipitation in each season. He divided the trends of secular change into the following three types: increasing (type 1), decreasing (type 2), and indefinite trend (type 3). Thg results are as follows:1. The June-July season. 1) Type 1 is distributed all over the country, except a few districts. 2) Type 2 is concentrated in the inland areas of Chubu and in Hokuriku.2. The August-November season.